What does a meaningful local chapter mean for the Greenpill Network? This time, we tried becoming sybils and discussed the meaning of sybils. QF can be effective or ineffective. A healthy, high-quality digital community can greatly enhance local social resilience, and effective public resource allocation will be the cornerstone of community governance.

In this article, I will focus on discussing why we operate sybil-attack at our related grant program, using actions to prove the effectiveness of quadratic funding(QF). To conclude, I believe that the 'local chapter' of the Greenpill Network is not easily developed into a mature, sustainable, non-profit action community, but is suitable as a stress tester for quadratic funding tools. Below, I will share the story step by step and analyze the reasons.

Context Introduction

First of all, Taiwan's FAB DAO is a community primarily engaged in non-profit web3 actions, closely interacting with the g0v citizen technology community project da0, with overlapping members and frequent joint events. As a member of the Greenpill Network, FAB DAO has a branch project called Green Sofa (Greenpill Taiwan), with about five regular contributors. The Greenpill Network, initiated by Kevin Owocki, aims to develop digital communities worldwide and advocate for sustainable web3 actions, with dozens of local communities, publications, and a regularly updated Podcast.

In November this year, we used four weeks to complete the first round of quadratic funding actions, carried out with the support of the Gitcoin Grant Stack tool and funded from Gitcoin Grant 18. The action met expectations, proving that a QF action can be fully executed in four weeks, creating reasonable funding and distribution results, as detailed in the article "FAB DAO Grant is Great: Results & Discussion of the Quadratic Funding Experiment". During the action, I mentioned in the Podcast 'Blocktrend' with MN Hsu (also a proponent and participant in public goods funding) that in this experiment we intentionally did not set identity qualification thresholds, i.e., no Gitcoin Passport score restrictions. We welcomed Sybil attacks to collect data and observe effects. Of course, there were no attacks, possibly because the 'Grant is great!' had only about $1,400 in public funds.

However, I did not expect to write a second article on quadratic funding experiments so soon. Because we conducted a sybil attack ourselves.

This was because Sejal Rekhan from the Greenpill Network, during her visit to Taipei, initiated a lightning QF event. December this year in Taipei is a web3 month, with events like Funding the Commons Taipei(FtC), Taipei Blockchain Week 2023(TBW 2023), DAO Taipei, NFT Taipei, etc. Sejal's event, called 'gov0/FAB DAO/dao0 Live Round,' lasted from 12/11 to 12/16, a total of only 6 days, coinciding with FtC and the day after the g0v hackathon, ending the day before DAO Taipei. The schedule was extremely tight, 20 days shorter than our own 'Grant is Great!'. Remember, the length of the schedule is very important, even more so than the amount of public funds, as it affects the quality of QF.

Quadratic Funding Process

The event started on 12/9 (Saturday). After attending the g0v hackathon in Hsinchu City, Sejal immediately participated in the FtC forum in Taipei and announced a QF in Taipei called 'gov0/FAB DAO/dao0 Live Round' (actually should be written as g0v and da0). She set up a workshop at the FtC venue to guide anyone interested in submitting proposals or donating. The public funds for this QF were $3,000, later found to be provided by the organizers of FtC, not the Greenpill Network, which was the first relatively unclear point, leaving willing collaborators unsure how to promote.

After Sejal announced this, members of Greensofa began discussing how to assist and felt anxious. This was because Greensofa had just completed a round of QF, and at the same time, another team had just completed a round of QV (Quadratic Voting) experiment 'FOSS and Open Data Playground' at g0v with about $13,000 to verify the effect of Retroactive Public Good Funding (RPGF). Another QF so soon was unexpected and its actual effect uncertain. First, this sudden announcement would certainly mobilize cooperators again, and second, such frequent QV/QF activities would also mobilize proposers and funders, not sure if it would reduce effectiveness due to crowding out or reduce participation due to cognitive fatigue.

In this process, the main collaborator assisting Sejal was Stanley from Greensofa, an experimental education teacher, handling translation, coordination, etc. After Greensofa's regular weekly meeting (attended by Stanley, g0v and FAB DAO participant Denken, and multi-organizational participant Swift), it was decided to help promote Sejal's event as much as possible. Technical assistance, like setting up wallets for new participation in funding, would be provided if needed, and if Chinese customer service was required, one could ask on FAB DAO Discord. The Greensofa team itself would not apply for projects.

Summarizing the above, in this round of QF, the promotions included:

1. Sejal's oral promotion at g0v hackathon and FtC

2. Sejal's promotion on g0v slack and Telegram

3. FAB DAO's promotion on Discord

4. There was no design for public page promotion this time.

This QF also did not set identity verification thresholds, with proposals manually reviewed by Sejal, similar to our own QF.

Quadratic Funding Results

From the results, there were 13 proposals, 9 of which overlapped with the 'Great Grant!' team already completed by FAB DAO, with 3 new proposals: a crystal bead workshop, a seed art workshop, and MozTW Space (Mozilla Community Space Taipei). The last one was Sejal's own proposal. Of these 13 proposals, only MozTW was from the g0v community.

The funding amount reached a considerable $5,135, more than the $3,000 public fund pool, with 24 funders and 90 funding instances. The Micetopia project (a free Taipei web3 event hat and game project) alone accounted for $4,638, details seen in the 'report card' image.

Unfortunately, there was not so much actual donation, mostly false funding from witch attacks, with the actual funding amount around $200 and only 10 donors. Judging from the funding behavior, these amounts were likely from the proposers themselves.

It can be said that the proposers were more enthusiastic than the funders in this QF.

The Story of the Sybil Attack

Based on only insiders participating, Gimmy Chang from Greensofa and also the initiator of Micetopia, and I decided to launch a sybil attack six hours before the end of QF. The attack was not difficult to execute, needing not many instances to be effective, and this attack did not cause collateral damage. We thought this fund deserved higher quality QF and could be used again with other public funds in the future for more effective Rounds.

Besides, we discovered flaws in the Grant Stack's system design and experimented with them.

We found that after funders transferred money, proposers' addresses immediately received the funds. So, we adopted a 'circular funding' process to generate high funding amounts.

For example, I donated $300 to the Micetopia project, and Gimmy immediately transferred $300 to my second address, which I then reused to donate to the same project, repeating the process. We eventually used 12 addresses to generate about $4,900 in funding. Based on QF rules, this actually created more than ten times the funding weight, so this round of attack would take over 90% of the public funds.

We also found some small flaws in the design of Gitcoin Grant Stack:

1. Funds are directly transferred to the recipient's address, making collusion easy. The risk of lowering the threshold is the inability to prevent collusion.

2. Using Metamask wallets encountered issues with proposing or funding.

3. The full amount in the wallet could not be used for funding. For example, with 50 DAI in my wallet, I couldn't donate the full $50 but had to give $49.99.

The final result was uncertain whether Sejal would receive this donation from FtC and whether this public fund would be distributed to each project as verbally promised. But we also verbally promised that if we received this fund, it would be used for public good purposes, such as more educational projects and the next round of 'Grant is great!' QF action.

Discussion of Results

Aside from technical flaws that are easy to fix, we are more interested in discussing community dynamics, the most important issue for the Greenpill Network in expanding territories. Because communities and territories are different, you can't claim ownership by planting a flag but must deal with intricate cultural issues.

Our core question is, what kind of international support can effectively promote local community development? Good methods will align global/local dual needs and avoid ineffective actions or easily misunderstood colonialist behaviors. To clarify, Sejal's enthusiastic lightning QF does not make us feel colonized, but I am concerned that when the Greenpill Network has substantial funds in the future and lacks a clever injection process, it will be easy to give people the wrong idea.

We believe QF will have different meanings for communities at different stages. For local communities initially understanding QF, the experiential effect is greater than actual funding; for experienced local communities, the community will care if QF meets the goal—effective resource distribution. After all, traditional grant rules and the capital market have difficulty effectively distributing resources related to public goods; otherwise, QF would not be so welcomed by Regens.

We think, in any case, the human resources of local collaborators are more important than public funds, as high-quality promotion, guidance, and planning will significantly affect the quality of public fund distribution. Without effective distribution, no matter how high the public funds, it will not persuade local non-profits to participate in QF actions, becoming a self-fulfilling prophecy. This argument can be drawn from comparing the 'Great Grant!' QF and Sejal's QF.

From the 'Great Grant!' QF, we can see that since the proposers were all from FAB DAO, the funding amounts were small, but aligned with everyone's independent value judgment, so these judgments together generated collective intelligence, distributing more effectively to different projects through QF. Here's an important phenomenon: as most donors were also from FAB DAO, they didn't viciously compete to increase funding amounts, nor did they try to conduct malicious collusive attacks.

In Sejal's QF, we can see a significant increase in individual funding amounts, tens of times higher, because the proposers did not know each other, lacking a 'courteous' atmosphere, and each tried to increase funding amounts to get more public funds. Although both events were QF, they produced completely different effects.

Why is this so?

I think it can be analyzed using the eight principles of public goods governance (Governing the Commons) by Nobel laureate in economics Olinor Ostrom:

1. Define clear group boundaries.

2. Match rules governing use of common goods to local needs and conditions.

3. Ensure that those affected by the rules can participate in modifying the rules.

4. Make sure the rule-making rights of community members are respected by outside authorities.

5. Develop a system, carried out by community members, for monitoring members’ behavior.

6. Use graduated sanctions for rule violators.

7. Provide accessible, low-cost means for dispute resolution.

8. Build responsibility for governing the common resource in nested tiers from the lowest level up to the entire interconnected system.

To me, the most significant difference in these two QFs is the clearly defined boundaries and actual needs of the local community. The same rules, even if executed by the same group, will produce different results with different narratives. Moreover, considering that QF still has a high participation threshold, this new mechanism benefits funders feeling the actual benefit of funding, thereby increasing motivation to participate, but the downside is vulnerability to sybil attacks. Balancing threshold settings and witch attacks needs to be determined based on the range and purpose of different QFs.

Therefore, we suggest:

1. Local collaborators are key because they can effectively influence other participants and ensure meaningful 'distribution' during QF execution, as defined by different local communities.

2. The opinions of local community collaborators, especially unpaid contributors, should be valued. This will reduce the likelihood of collusion and improve distribution quality.

3. Each round of QF should have a complete promotional cycle and guidance setup.

4. Before each round of QF, potential participants should be clearly defined, whether in a DAO, a region, or at an international level.

5. Each round of QF should preliminarily investigate the actual needs of potential participants, including proposers and funders, to enhance their long-term participation or regular funding motivation.

6. Greenpill local communities are currently most useful not as sustainable communities (thinking so easily leads to bubbles) but as stress testers for Grant Stack.

Conclusion

Healthy, high-quality digital communities can greatly enhance local social resilience, and effective public resource distribution will be the cornerstone of community governance. No single mechanism design can universally fit all cultures, but we know that under sufficient discussion and experimentation, different local communities will slowly find distribution methods that best suit their interests.

We discovered these important community values through sybil attacks, providing them for all Greenpill Network reference.